一种针对水泥回转窑故障诊断的贝叶斯网络模型

为了实现水泥回转窑的故障诊断，采用贝叶斯网络建立了水泥回转窑故障智能诊断模型。在模型建立 过程中，提出了一种基于数据样本、不依赖先验知识的贝叶斯网络结构学习改进算法。在利用改进结构学习算法 建立诊断模型贝叶斯网络的基础上，利用MLE算法和变量消除法完成了模型的参数学习和诊断推理。为了验证 水泥回转窑故障诊断贝叶斯网络模型的准确率以及可行性，利用现场数据进行了大量的测试实验。

A Fault Diagnosis Bayesian Network Model for Cement Rotary Kiln

In order to realize fault diagnosis of the cement rotary kiln, Bayesian Network was used to establish the model of rotary kiln intelligent diagnosis. In the process of building model, an improved Bayesian Network structure learning algorithm was proposed. The improved algorithm requires dataset but doesn⁵1 rely on prior knowledge. Based on the Bayesian Network established by the improved algorithm, Maximum likelihood estimation ( MLE) algorithm and variable elimination method are used to complete parameter study and diagnosis reasoning. To testify accuracy and feasibility of cement rotary kiln fault diagnosis Bayesian Network model, plenty of experiments were conducted with field data.